List your company

Mathematical Reasoning on AIME25 (Accuracy, Efficiency k)

46.7Accuracy (%)Qwen3-4B-Inst-2507
-1.86810.74123.3535.959Feb 5, 2026

Evaluation results

MethodLinks
Qwen3-4B-Inst-2507Chat Template=Off, Decoding=Baseline, Step=0, k=12026.0246.71
Qwen3-4B-Inst-2507Chat Template=Off, Decoding=ConfAdapt, tau=0.9952026.0226.72.3
Qwen3-4B-Inst-2507Chat Template=Off, Decoding=Static, k=12026.0223.31
Qwen3-4B-Inst-2507Chat Template=Off, Decoding=ConfAdapt, tau=0.952026.0223.32.8
L3.1-8B-MagpieChat Template=On, Decoding=Baseline, Step=0, k=12026.0201
L3.1-8B-MagpieChat Template=On, Decoding=Static, k=12026.0201
L3.1-8B-MagpieChat Template=On, Decoding=ConfAdapt, tau=0.92026.0205.5